Académique Documents
Professionnel Documents
Culture Documents
size, TOEIC
scores, and
testwiseness
Masaya Kanzaki
Kanda Institute of Foreign
Languages
Reference data:
Kanzaki, M. (2010). Vocabulary size, TOEIC scores, and testwiseness. In A. M. Stoke (Ed.), JALT2009 Conference Proceedings. Tokyo: JALT.
The relationship between vocabulary and English proficiency tests was investigated among 31 learners.
They were given the Vocabulary Levels Test (120 questions) and a TOEIC practice test (200 questions),
and their scores were compared. The correlation between the two tests was .64. However, when the
subjects were divided into two groups based on the number of times they had taken the TOEIC, the
figure for the expert group (n = 15, exam taken = 8 times or more) was .76 and that for the novice
group (n = 16, exam taken = 6 times or fewer) was .49. This paper examines the difference between
the two groups in relation to testwiseness, a respondents capacity to use knowledge about a test itself,
in spite of what the test is supposed to measure, in order to gain a high score.
31120TOEIC200
20.64TOEIC2
1580.7616
60.492
huge amount of English vocabulary, EFL learners have to learn a lot of new
words when acquiring proficiency in the language. The development of a learners
overall English ability often correlates with growth of vocabulary knowledge, and vocabulary size is reflected in a learners English proficiency test performance. Nation and Meara
(2002) observed that there is a relatively close relationship between how many words you
know, as measured on the standard vocabulary tests, and how well you perform on reading
tests, listening tests and other formal tests of your English ability (p. 50).
iven the
Several studies have explored this close relationship. Qian (1999) compared the Vocabulary
Levels Test scores with the TOEFL reading comprehension section scores of 74 ESL students in
Canada and found that the two tests correlated well at .78. Similarly, Beglar and Hunt (1999)
compared the 2000 Word Level Test scores with the TOEFL scores of 496 EFL students in Japan,
and the correlation between the two tests was .71.
These results agree with what is often observed in the classroom. When students cannot
answer practice test questions correctly, it is often a lack of vocabulary hindering their success.
PREVIOUS PAGE
NEXT PAGE
FULL SCREEN
740
Under 500
500545
550595
600645
650695
700745
750795
800845
850895
900990
Note: Based on highest scores. One of the participants had never taken the TOEIC.
10
11
12
13
14
15
16+
2029
3039
4049
5059
6069
70+
12
JALT2009 CONFERENCE
PROCEEDINGS
ONLINE
PREVIOUS PAGE
NEXT PAGE
FULL SCREEN
741
Method
A vocabulary test and a TOEIC practice test were given to 31
participants, and the scores of the two tests were compared.
1.
business
3.
horse
Participants
4.
Materials
As a measure of vocabulary size, four levels of the Vocabulary
Levels Test were used (30 questions per level, 120 questions in
total). As a substitute for the actual TOEIC, 200 questions from a
practice test book were used.
Each level of the test has 10 clusters of 6 words with 3 definitions, which makes a total of 30 questions per level. An example
of a cluster is shown below.
JALT2009 CONFERENCE
PROCEEDINGS
ONLINE
2.
5.
6.
clock
pencil
shoe
wall
Test takers are instructed to match the words on the left with
the definitions on the right. Schmitt, Schmitt, and Clapham
(2001) reported the Cronbachs alpha reliability values for the
2000, 3000, 5000, and academic word levels of Version 1 as .920,
.929, .927, and .958, respectively. Therefore, this vocabulary test
is reliable.
PREVIOUS PAGE
NEXT PAGE
FULL SCREEN
742
Listening
Reading
Part
Task
# of Qs
10
30
30
30
40
12
48
Procedures
Two practice tests with a total of 400 questions in the practice
test book were used during the four-day intensive course.
To provide the same number of questions on each day of the
course, the two tests were divided into four half tests and the
participants answered 100 questions each day (i.e., the first
half of Test One on the first day, the second half of Test One on
the second day, the first half of Test Two on the third day, and
the second half of Test Two on the fourth day). In addition, the
JALT2009 CONFERENCE
PROCEEDINGS
ONLINE
Results
Descriptive statistics
Table 5 shows the descriptive statistics for the Vocabulary Levels
Test. The mean of the academic word level is between those of
the 2000 and 3000 levels, which agrees with the results reported
by Schmitt et al. (2001).
2000
3000
5000
30
30
30
30
27.58
24.42
21.29
25.48
98.77
Range
16
17
20
20
67
SD
3.29
4.24
5.20
Academic
3.71
Total
120
14.57
PREVIOUS PAGE
NEXT PAGE
FULL SCREEN
743
Listening
Reading
TOEIC-Total
2000
.28
.72
.55
3000
.36
.71
.61
5000
.42
.67
.61
Academic
.27
.62
.49
Section
VLT-Total
.39
.76
.64
Listening
100
Reading
100
Total
200
70.39
74.71
Range
49
48
83
SD
12.25
11.51
21.17
.87
.87
.92
3.97
3.77
5.51
K-R21
SEM
145.10
Correlation
Table 7 shows the correlation coefficients between the two tests.
The reading section of the practice test correlated well with the
vocabulary test at .72 with the 2000 level, .71 with the 3000 level,
.67 with the 5000 level, .62 with the academic word level, and
.76 with the total of the four levels. The figures for the listening
section, however, were much lower at .28, .36, .42, .27, and .39,
respectively. The correlation coefficient between the totals of the
two tests was .64, which is statistically significant (p < .01).
JALT2009 CONFERENCE
PROCEEDINGS
ONLINE
PREVIOUS PAGE
NEXT PAGE
FULL SCREEN
744
2000
3000
5000
Academic
Total
Novice
Expert
Novice
Expert
Novice
Expert
Novice
Expert
Novice
Expert
27.75
27.40
25.00
23.80
20.94
21.69
25.69
25.27
99.38
98.13
Range
12
16
13
17
18
20
13
19
43
67
12.91
16.13
SD
2.73
3.79
3.95
4.45
5.24
5.13
3.23
4.14
Listening
Reading
Total
Novice
Expert
Novice
Expert
Novice
Expert
68.88
72.00
75.56
73.80
144.44
Range
45
44
37
46
70
83
SD
12.45
11.82
10.73
12.23
20.07
22.27
145.80
Listening
Novice
Reading
Expert
Novice
TOEIC-Total
Expert
Novice
Expert
2000
.00
.54
.48
.88
.25
.77
3000
.23
.53
.55
.84
.44
.74
5000
.37
.47
.64
.72
.57
.65
Academic
.01
.53
.50
.71
.27
.67
VLT-Total
.22
.56
.65
.85
.49
.76
JALT2009 CONFERENCE
PROCEEDINGS
ONLINE
PREVIOUS PAGE
NEXT PAGE
FULL SCREEN
745
Discussion
Possible reasons for weak correlations
As shown in Table 7, the reading section of the practice test
correlated well with the vocabulary test while the correlation between the listening section and the vocabulary test was
weak. One possible reason for this is the Vocabulary Level Test
measures vocabulary knowledge only in written form. Since
the listening section requires test takers to recognize words in
spoken form, an aspect of vocabulary knowledge not measured
in the Vocabulary Levels Test is relevant in the listening section
of the TOEIC.
JALT2009 CONFERENCE
PROCEEDINGS
ONLINE
Word list
Listening
Reading
1000
86.56
74.40
2000
94.07
84.93
3000
95.89
88.63
4000
96.80
91.36
5000
97.14
92.60
PREVIOUS PAGE
NEXT PAGE
FULL SCREEN
746
JALT2009 CONFERENCE
PROCEEDINGS
ONLINE
Testwiseness threshold
The testwiseness ceiling is a threshold for fair score comparison.
Until learners have come to this ceiling, their TOEIC scores
cannot be compared fairly. This is because those with low levels
of testwiseness have an unfair disadvantage; their scores are
likely to be lower than those with higher levels of testwiseness,
even when English proficiency levels are the same. In other
words, when learners have reached the testwiseness threshold,
their TOEIC score begins to work as an indicator of their English
proficiency.
Conclusion
This study revealed that vocabulary size and TOEIC scores
correlate to a moderate degree, as the correlation coefficient
PREVIOUS PAGE
NEXT PAGE
FULL SCREEN
747
Bio data
Heatley, A., Nation, I. S. P., & Coxhead, A. (2002). RANGE (version 32)
[Computer program]. Retrieved January 12, 2007, from http://www.
vuw.ac.nz/lals/staff/Paul_Nation
Qian, D. D. (1999). Assessing the roles of depth and breadth of vocabulary knowledge in reading comprehension. Canadian Modern Language
Review, 56(2), 282-307.
References
Beglar, D., & Hunt, A. (1999). Revising and validating the 2000 word
level and university word level vocabulary tests. Language Testing,
16(2), 131-162.
JALT2009 CONFERENCE
PROCEEDINGS
ONLINE
PREVIOUS PAGE
NEXT PAGE
FULL SCREEN
748